THE BOOK--Playing The Percentages In Baseball

Btw, pretty convenient to presume the worst of me and the readers here, and then make that your conclusion.

The option would have been easy: engage us, and then be appalled if we lived down to your expectations.

One of the commenters at that site, sportsphd, suggested the following course of action:

---snip---
As Bradbury mentioned, if you want to challenge the Hakes and Sauer model, you submit a detailed refutation to either the authors or the journal. At that point, the original authors are expected to respond.
---snip---

It seems to me that you have more than enough material to write a formal refutation in the journal. Why not do it?

Millsy’s other claim:

--
The current state of The Book Blog reminds me of sitting in MBA economics classes at the Business School here at Michigan. Without understanding of how models are simplistic in order to explain expectations of markets, arrogant students consistently chastise the professor because of a single example they ran into at work. The professor’s response is always, “Well, of course there are anomolies, but on average X happens” as if he was waiting for it. This statement just doesn’t get through to people for some reason, despite its generality. The thing that blows my mind about this entire problem is that Sabermetricians believe they are critical thinkers. They are people who ran into others ignoring their opinions for years or criticizing their silly ‘statistics’ based on small sample sizes for years. Yet, the arrogance continues to blind minds to the fact that economics and Sabermetric study are so interrelated that ignoring the basic economic principles can be counterproductive in progressing the science.
--

Couldn’t his cult claim of Tango be applied to JC for the “sabernomics” crowd? If Millsy wasn’t following JC blindly, wouldn’t he come here and make his defense?

As the author of the comment quoted @2, I think that is the way to go with this.  Hakes and Sauer seem to have an awful lot hanging on a very small n.  With a sample size that small, as you point in your comments about the coding of the free agent variable, the coefficient point estimates could be wildly off.  It seems that those estimates need some sort of confidence interval.  Without that measure, it is hard to know what to do with the coefficients at all.  Given that most economics literature reviews and thus scholarship will be based on searches of economics journals and not the internet more broadly, the best way to bring out better economics research in the future is to correct the literature in that venue.

John, why is it Tango’s responsibility to submit something to a journal or the authors?  Advanced statistical analysis has been a collaborative effort for years and hopefully will remain that way.  Tango definitely has the material necessary to submit his work, but Tango handled this the way the sabermetric community has handled disagreements for over a decade now.  Why should it change so these two academics have to recognize the flaws in their work?  If they don’t want to be a part of the sabermetric community then that tells me they’re too fearful to have their work dissected by that community.  I hope Tango does submit his work here, but that is not how work has been challenged in the sabermetric circle.  Quite the opposite.  Why does Tango have to play by their rules and why can they ignore a process that has been ongoing in this community for a long time?  It seems to me if some academics want to research and write about baseball stats when they are unqualified to do so that Tango shouldn’t have to go through any hoops to prove them wrong.  He has already done that and whether he submits his work doesn’t change that simple fact.  If the authors wish to keep the same stance, so be it.

David/5:

It’s not a question of the “rightness” of submitting to a journal. It’s a question of efficacy. I’m not arguing that Tango has some responsibility to post his thoughts in a journal. However, I feel that it would be an effective way of accomplishing the intended goal: to point out the errors in the methodology and to enter into a debate about them.

FWIW, I’ve also suggested that Bradbury and Millsy stop by here, given the open dialogue this site is known for.

Hakes and Sauer no doubt spent a lot of time getting their article written, edited, and peer-reviewed.  They probably had to make a whole bunch of changes at the request of the referees.  The refereeing process takes a long, long time.

For anyone to suggest that Tango has to do the same thing, to rebut the article, is silly.  For one thing, Hakes and Sauer got paid handsomely for their work—Tango would have to do it for free.  That’s churlish. 

Second, the standards for academic publishing probably require a strong background in the theoretical academic side of the subject, even if that knowledge is not relevant to the rebuttal. 

Third, if you describe and praise an article in a blog, what the hell are you doing telling people not to criticize the article in a blog? 

Fourth, a rebuttal is often much simpler than the original hypothesis.  You can spend years perfecting an academic article that shows that no baseball player can hit his peak after 33.  It takes three seconds for me to show you Barry Bonds. 

Fifth, if the “rebutter” is right, and can prove it, the original author should be happy to hear it as soon as possible, by any means available.  Even the most vain academic would agree.  If Dr. X mistakenly adds 1+1 and gets 3, he’d be happy if I told him so he could fix it.  There’s no way he’d tell me to shut up and go submit a rebuttal.

So “if you disagree, you should get your rebuttal published academically instead of arguing on your blog” really means, “I still think you’re wrong and I’m sick of arguing.  This is my way of trying to get you to shut up ... and, besides, maybe a referee will be able to knock some sense into you.” I think it’s more polite to say it that way, instead of implying that Tango is somehow being rude by making the argument outside of academia.

All fair points, Phil.

Shawn/8: Well, I, too, would think it were cool if Tango wrote a rebuttal and it were published in that same journal.  I’m not sure how much good it would do ... In JQAS 2.3 article 5, Roland Beech wrote a negative review of “The Wages of Wins.” That’s an academic journal.  But I don’t think it convinced anyone, really.  In JQAS 2.4.4 article 4, David Berri really laid into Beech.  Berri didn’t seem to have a whole lot of respect for the fact that Beech’s review passed peer review. 

Remember: the book was not peer reviewed.  The review was.  If academics were consistent, they would never cite TWOW, since the only peer-reviewed article was against it.  At least, they wouldn’t have cited it for the three months in which it hadn’t been rebutted.  I doubt it happened that way.

And, suppose that Tango DOES get his rebuttal published, making exactly the same arguments that he did on this blog.  Are JC and Millsy now going to say, “Geez, I guess Tango was right all along?”

The whole “publish in a journal” argument from academics is a red herring, like when creationists say “more research is required” before accepting evolution.  It’s something you say to make it look like you’re furthering the debate, when what you’re really trying to do is cut it off.

@7:

Despite the interesting attempt at mindreading, you seem to miss my point completely.  In no forum have I suggested that Tango should not publish his corrections wherever he so chooses.  Instead, I think this is a question of making sabermetric research as good as it can possibly be.  While Tango may not be able to get an article through peer review because of the unique requirements of economics journals, that has no bearing on the question of whether a rebuttal should be submitted.  Maybe, instead, someone with more experience with economics journals should try to use Tango’s work as a stepping stone to further consideration of the topic broached by Hakes and Sauer.

It seems to me that sabermetric research has better prospects if it can attract more possible eyes and thoughts on any given question.  Focusing simply on blogs by definition will narrow who will attempt to answer the questions we all care about.  In other words, I think shawn @6 is correct.  The question is efficacy.  What gets us the best research about baseball.  I think that challenging this argument in academic circles will achieve that goal.  If you disagree, please disagree with that point because that is the point I am trying to make.

John/10:

I’m arguing more that it’s wrong for academics to EXPECT or DEMAND that Tango submit.  I’m not really arguing that it would do no good.

In 9, I argue that publishing won’t do anything to change the minds of anyone disagreeing with Tango who is demanding that he be published.  Those guys don’t care if he’s published—their minds are made up.  But I agree with you that it would make a difference to people new to the debate, as well as quiet some of the people (JC) who are now able to dismiss him because he is only a pseudonymous blogger.

It’s all a matter of whether it’s worth the time and effort.  Academics get paid well for publishing articles.  We have to do it on our own time and money.  As a one-time shot, well, maybe.  But if you expect sabermetricians to regularly spend fifty unpaid hours preparing a journal article that won’t be published for a year, when they could write it and have it up in 90 minutes on a blog ... well, that’s just not going to happen.  Nor should it.

Wondered why I suddenly had blog visitors besides my mother and my girlfriend.  Now I see.  And mostly all fair points.

However, to claim I do not try to engage here is also misleading.  But no, I am not a commenter who participates as much as those like Phil, Guy, Rally, Colin and Patriot.

Shawngoldman/2,
If I were a journal editor, I would absolutely welcome it, despite what other researchers may think.  Thanks for your comments on my site as well.  While Tango does point out my chastising, I also do state that I think this site is FILELD with extremely intelligent people who I do believe have much more skill and experience than I do.  Unfortunately, as I plan to include in a later post, I also agree that there is an arrogance on the other side of things.  My original interest in the topic of analyzing sports came from places like By The Numbers and blogs here online, so while I often dissent, I really try sympathize with both sides (but given my current status, more so with the academic side). 

JD/3,
I think it’s an exagerration to say I follow JC blindly.  And I am here defending my opinions.  I have often inquired about and questioned his valuations presented at his site in the past.  I’ve also stated here that his model (which he now ackknowledges) seemed to stretch the idea of the amount of revenue players were subject to getting.  I do enjoy much of his analysis, but I’ve seen plenty of things I disagree with.  I also understand the dislike of his responses, and empathize with that.  But often times it seems that there is outrage over fairly rational points made simply because it’s JC Bradbury.

My point about the MBA class was simply toward the awfully large dissent against the ‘Myths’ by Bradbury.  The points as a whole were extremely general, and consistently anecdotal evidence was presented.  I don’t necessarily fully agree with everything he says, and I don’t think I’m alone.  But the response was as if the points made were outrageous, when they were simply an assumption of what the average rational actor would accomplish.  I think this reaction was the main catalyst to my post.

My former problems with replacement player were stated here, and Tango argued his point very well to the point that I realized some of my thinking was completely wrong.  While I disagreed with some of the original ideas I had about the topic, I did acknowledge that much of it had its foundation in a good economic place.  I acknowledged my error on my own site linked above.

Phil/7/9/11,
I don’t think it’s rude one bit to make an argument outside of academia.  The original post that I saw did use extrapolated values.  However, I also think there is plenty of things to question in the study.  Tango makes some later points that do interest me a little more and I agree could be problems.  I definitely would not just appeal to authority in this situation, as I hope no rational person would.  I think the problem I had was that the point of the study was misrepresented a little bit, given it wasn’t a salary model.  But I do think there were some problems with it.

Not once did I defer to the peer review process.  I’ve seen a plethora of crap in certain journals I won’t name, so I think that’s in itself a red herring to anything I said.  While I agree it’s a great process to have, by no means do I agree that its ability to weed out junk is beyond criticism.  I also know specific people in my area that value the discussion on your blog very highly.

I agree with you 100% on your last point in 11.  I honestly do not expect someone outside of academia to use their time to publish exclusively in academic journals.  Easy access can be very productive (as much of the history of sabermetrics has shown), but also comes with a sense of responsibility.  This is also why I ‘brain dump’ at my site.

Tango,
I guess the response should mostly be to you.  You seem to have taken things personally.  My conern is not with your analyses or with the fact that you present your opinions here on academic work (in fact, as someone who is training to work in the academic world, I value the input that you could definitely provide).  I TRULY hope that the upcoming generation in sports, economics, and quantitative analysis can bridge gaps that seem to have been made.  This was the point of my post, which seemed to be lost in translation.

The concern I have, and I think many I know of in academia have, is with those reading the site who will ultimately be using it to appeal to authority, and sometimes do the exact thing Bradbury does when arguing about how we should trust economists to the core.  A point that JC makes about using work by PhDs--which I think he overstates considerably.

I did not make myself clear, I guess, with my post as to my difficulty with the community and its medium.  The post comes off as emotional in a way that was generally sparked by comments like “fuck you” or making generalities about economists as bad scientists (a subject which I have stated my concern with on The Book Blog before).

What I wrestle with is the public nature of the internet (and I think this is a problem that many academics I know have).  While blogging is a GREAT tool, its downfall is that many things can be presented in a persuasive manner to those reading it who really do not have the ability to critically analyze it themselves.  I think this was highlighted VERY strongly by the response that Bradbury mentions in his interview, where that responder states right in the interview comments that he doesn’t understand multiple regression, but that since you think it’s wrong, that’s good enough for him.  That’s NOT to say you’re your analysis is wrong, but the beginning of it did concern me.  And I, too, think there could be problems with the study.

Many I’ve come across at sabermetric sites seem to point out that it is problematic when deferring to an Economist, but not when referring to a sabermetrician.  I think it’s problematic on both fronts.  It is arrogant from either side.  Does that happen with everyone?  No, so perhaps I made generalizations.  But I don’t think you can deny this happens.  There is a myopic view about the issue on BOTH sides.  Rereading my post, I perhaps have pushed this myopic view myself on one side and not the other.

A number of the posters here do in fact make interesting points, which is why I come here.  And also why The Book Blog is linked on my sidebar.

I think my discontent was with the outlash toward points such as the idea that buying low and selling high is not a reasonable strategy to have as a general manager—this was the catalyst for the MBA student analogy.  I DO think you present it well in the blog as a “what do you think” post.  GMs are not all really, really smart guys—which is why they hire Tango and other statisticians—and there is information asymmetry so this is a strong assumption. But not one that is “just plain wrong”.  If it were the case that a GM was selling low on a consistent basis, he would not be keeping his job.  I don’t think anyone here really thinks that’s a ridiculous statement.  Matt Swartz tried to make these points, but others seemed to find anecdotal points to refute the points he made, or yes, just ignore the fact that they are general assumptions.  Not every poster did, but some definitely did.

Strong opinions can often be persuasive to those who don’t have a full understanding of the problem, and writing off the type of claims that Rosenthal makes ignores the fact that this exists.  Both Rosenthal and Bradbury acknowledge great strides made by the sabermetric community, and in fact single you out, Tango, as a leader.  I would have to agree, and find the detail in The Book fascinating.

Did millsy ever refute the actual findings in that long response? Can someone debate Tango on the study instead of debating Tango on debating the study?

Post your response, dude; I don’t think it will fall on “deaf ears.”

Sam/13,

I stated my concern with the original post on the subject and extrapolated values above.  I also stated that Tango went further with it, and I find that interesting.  However, my original post was before I saw these followups, which I also state above.

Another thought on this:

Would it be possible to use WAR/linear weights to get at the question Sauer and Hakes are trying to address; namely if OBP has been valued more since the publication of “Moneyball?”

I’m thinking we could separate the RAR resulting from a player’s OBP and the RAR resulting from a player’s ISO. If we were to do that, and then applied the same methods used to get at $/WAR to the OBP/ISO components of WAR, we may be able to verify or refute the conclusions in the original paper using separate methodologies. Then, regardless of complaints of methodology we’d be able to answer the fundamental question that is at stake here.

Millsy -

I stated my concern with the original post on the subject and extrapolated values above.

On multiple occasions Tango has taken non-extreme values in plugged them into the regression model.  The results still don’t make sense. 

He has outlined his problems with the study using normal values, common sense, and other necessary things used to review a study.

http://www.insidethebook.com/ee/index.php/site/article/going_through_the_sauer_hakes_moneyball_data/
http://www.insidethebook.com/ee/index.php/site/article/the_sauer_hakes_moneyball_spreadsheet/

Can you, or JC, please address those points instead of A) criticizing the few occurrences in which extrapolation occurred, or B) criticizing the way in which Tango criticized the Sauer-Hakes model.

Millsy, I appreciate your thoughtful response here.  As far as I know, you are the first person from academia to acknowledge that the collective sabermetric frustration with the approach by the economic academics on this topic is not unfounded.

Maybe, instead, someone with more experience with economics journals should try to use Tango’s work as a stepping stone to further consideration of the topic broached by Hakes and Sauer.

That would be wonderful. 

It seems to me that sabermetric research has better prospects if it can attract more possible eyes and thoughts on any given question.  Focusing simply on blogs by definition will narrow who will attempt to answer the questions we all care about.  In other words, I think shawn @6 is correct.  The question is efficacy.  What gets us the best research about baseball.  I think that challenging this argument in academic circles will achieve that goal.  If you disagree, please disagree with that point because that is the point I am trying to make.

In terms of efficacy, I don’t think anything can beat blogs.

***

You would think the authors of the paper would be interested when I bring up factual points (like there could not have been 14 free agents in 2001, when there was 137 or whatever in 2009).  Nope, there was no “wow, really?”.  I was disappointed by the reaction frankly. 

The Tejada thing (signs for 6/72, front-loaded, and authors mark his first year salary at 5MM instead of 12MM) had the authors reply that they were true to their source.  Well, yeah, but that’s besides the point. 

***

However, to claim I do not try to engage here is also misleading.

I did not say you didn’t engage here.  I said you didn’t engage on this particular issue.  Indeed, that’s what you said, that you purposely did not engage here.  And I’m saying you should engage me here, since I have an open forum.

***

I think this reaction was the main catalyst to my post.

I pretty much agree with you on the JC/myths thing.  And if you look back, I was pretty quiet about the whole thing.  I thought his article was pretty ho-hum, and had pretty much no comment on it. 

***

I think the problem I had was that the point of the study was misrepresented a little bit, given it wasn’t a salary model.  But I do think there were some problems with it.

Even if it is not a salary model, that it spit out such impossible numbers would at least point to an issue of bias somewhere.  And, having a free agent infielder marked as making 5MM instead of 12MM (among many other problems) would seem to be a big deal to me.

***

You seem to have taken things personally.

Incorrect.  I am personally unfazed.  I am professionally (or bloggilly peeved).

Gotta go play with my kid now… I’ll reply back after he goes to bed.

Nick/16,

I have twice acknowledged here that Tango went further with the analysis.  The idea about extrapolation was of concern before, but not with the additional things he provided.  Which is why I haven’t pressed on the issue.

Thank you for your response, Tango.  Reasoned, and deserved.  I agree that things like the Tejada concern are real, and I also think this is something that will come up in the “Data Issues” special edition of The Journal of Sports Economics.  Consistent definition of variables is something I see often missing when analyzing sports in academia (and other places).  There is so much data, which makes it attractive, but also very difficult to sift through.  Some are better at it than others.  I’ve run into a lot of ‘salary’ vs. ‘contract’ problems with my own data collection.  I am extremely happy that you took the time to respond, whether it be in a positive or negative light.

What I wrestle with is the public nature of the internet (and I think this is a problem that many academics I know have).  While blogging is a GREAT tool, its downfall is that many things can be presented in a persuasive manner to those reading it who really do not have the ability to critically analyze it themselves.  I think this was highlighted VERY strongly by the response that Bradbury mentions in his interview, where that responder states right in the interview comments that he doesn’t understand multiple regression, but that since you think it’s wrong, that’s good enough for him.

Here’s how I see it:
1) Hakes and Sauer put out a model that explains how baseball players get paid in regards to various factors
2) Tango points out that the back data doesn’t reflect how salary really works, and the results don’t make sense (infielders are worth less than outfielders, etc.)
3) People complain that Tango doesn’t understand the concept of regression and applying it outside a sample / People complain that Tango is being blindly followed
4) Posters on this blog point out that Tango’s concerns aren’t being met

I’d wager this is how those on the academic side are seeing it:
1) Hakes and Sauer put out a model that explains how baseball players get paid in regards to various factors
2) Tango puts in absurd numbers to the models to show it doesn’t model reality, and clearly doesn’t understand how regression works
3) Through various other blogs/interviews/etc. the academic community tries to reach out and explain to Tango why he’s wrong
4) Rabble on Tango’s blog start throwing around vulgarity and following him rabidly despite not properly understanding the concepts involved

Yes, there are people on this blog who blindly follow tango. And there are plenty of people who blindly follow JC Bradbury, or any other public figure as well.

Let’s forget about those people—they don’t raise the level of the discussion.

There are also a lot of people on both sides with open minds who want to get at the heart of the matter—how the heck do we value players?

Tango takes a baseball-first approach, looking at actual production and then figuring out how much that production seems to be worth. Bradbury (and other sports economists) seem to go from salary first, and then try to figure out what they are paying that salary for.

Both approaches have merit in the sense that they should both end up matching in the end.

What isn’t good is if tango’s approach consistently turns up players who should be making $20 million/year, but are really only being paid $5 million. Or if models like Bradbury’s say that Francoeur was worth millions despite producing less than freely available talent.

Of all the models I’ve seen made by economists, none have been anywhere near as valuable in predicting reality as Tango’s have been. So I’d have to side with Tango so far.

So if you want to raise the level of debate, then address what Tango is actually saying. Don’t make this some semantic argument about whether he understands how economists use regression or not—that doesn’t help us get anywhere nearer to the heart of the matter.

If regression works, and the Hakes & Sauer model doesn’t, it means that the Hakes & Sauer model has faults. Don’t try to blow rainbows up my ass and tell me that it works and we just don’t understand linear regression the way it’s used here.

What are the odds that an individual lacking specific “academic credentials” would get past an editor to even get to engage the peer review process?

It’s disingenuous for JC to suggest such a course of action. At best it’s arrogance.

Trying to prove that Moneyball increased the price of OBP is a fool’s errand.  The only place you can really look is the free agent market, and there just aren’t that many free agents every year to give you enough of a sample.  The player salaries are affected by other things like position, defensive ability, age, scouting opinions.  Some of this information isn’t even available.  You can control for the rest with regression, but I don’t think any study could sift through all these factors, given the small sample, and come up with any conclusion that falls outside an error range.

Rally,

“You can control for the rest with regression, but I don’t think any study could sift through all these factors, given the small sample, and come up with any conclusion that falls outside an error range.”

That’s essentially what I’m proposing. Look at the things we can measure on the same scale (WAR/RAR) and see how much each (OBP in particular) has contributed to the value of a player. In such a study, you could say that to within some certainty (in $/obpWAR), an increase in the market value of OBP skills did/didn’t occur. It seems that would be a better approach the problem than the one by Sauer and Hakes, and would solve many (but not all) of the problems raised about the study.

Rally/22,

I agree that trying to truly show a shift in valuation over that small course of time is difficult.  While I think Moneyball may have sped up the diffusion of using other metrics, I think we were going in that direction to begin with.  So it’s really difficult, you’re correct, to make large conclusions off of that data.  Even if it was the reason for a market correction, I suspect it wouldn’t be grasped by teams dramatically from one single offseason to the next.

As someone who is enjoying the back-and-forth the last week or so between JC and Tango and the whole academia v. saber discussions I’d just like to drop my 2 cents that I feel a lot of the academia sources I’ve read (and I don’t read nearly as much as I do sabermetric sites/blogs/sources) are guilty of many of the things (in regards to tone/snark/"club-like" qualities) that they accuse the sabermetric circle of having.

I know that’s not entirely related but I also just wanted to subscribe to this thread to get updates on new posts.

Shawn/John:  Phil has raised most of the main reasons that it isn’t practical for sabermetric researchers to use academic journals as a venue for criticizing academic work.  There’s just no incentive equal to the hassle/barriers.  But the ideas and information are out there, freely available.  And we’re about 100x faster than the academic process:  with the recent Kovash-Levitt paper, it took us about 3 days to uncover multiple flaws and point the way to a correct analysis.  The authors can incorporate the criticisms (when valid, of course) and revise their work. 

Unfortunately, that doesn’t seem to happen much.
And if they won’t, then those who do have the right incentive should write the journal articles:  young academics like Millsy who need to make their mark.  For example, if a young economist wanted to expose the enormous flaws in Dave Berri’s wins produced metric, everything you need to know is already available on the web.  The problems with the Hakes Sauer paper were first pointed out years ago.  Or someone could write a good paper explaining how Bradbury’s and Fair’s aging studies are wrong.  Why doesn’t this happen?  I don’t know.  I assume at some point ambitious young economics grad students and junior faculty will figure out we’ve done 2/3 of the work, and will run with these ideas. 

But if young economists aren’t willing to do the work and jump through the necessary hoops, why should the saber community?  After all, we aren’t relying on these flawed works—they’re not really a problem for us, they’re a problem for economists. 

Millsy:  The problems with Hakes-Sauer are not limited to a possibly flawed salary model. It’s much more fundamental.  First, they never bothered to assess whether the mispricing of OBP, according to their own model, accounted for the A’s success as they claim.  In fact, this can’t possibly be true, because the values you get for the A’s players under the “wrong” pre-Moneyball model and the presumably “right” 2004 model are nearly the same.  The mispricing, even if you believe their model, has trivial consequences in terms of wins and losses.  Second, the correction they find in 2004 cannot possibly have happened.  The large majority of 2004 salaries were either set earlier in multiyear contracts or by an arbitrator relying on pre-2004 precedents.  There may have been a small underpricing of OBP.  But the main story they tell—that Oakland won by successfully exploiting that inefficiency, and that the inefficiency vanished quickly in 2004—is simply not correct.

Academic journals were written by academics for academics (in general), and paid for by the institutions that hired those academics.

I would much prefer to see an aggregating website where academics jot down their ideas (good or bad) on a nice open forum, in plain English, so that people interested in that topic can offer their comments.

We are seeing so much more cross-discipline stuff, that it would be nice to have people knowledgeable about both sides contribute to making a more accurate article.

So the economists would post their models, and we would critique them. We would post our valuation models, and they would critique them. And as a result, we’d get a nice model that combines the sabermetric focus on what generates wins with the economic perspective on how the configuration of major league baseball turns those wins into salary figures for players.

Wouldn’t that be swell?

Unfortunately, it sounds like the academic community isn’t anxious to embrace blogs and amateurs as a complement to peer review. And that’s a pity.

Hopefully it will change, but this whole exchange reminds me of what a businessman-turned-teacher friend of mine said about academia, “People who work in education for an extended period tend to lose perspective. They get hired to be smarter than everyone around them and to correct them, they can’t be fired, and they rarely get challenged or critiqued directly.”

Guy/23:

I agree with everything written here, particularly the bit about the blogs being more efficient than academic journals. I said this in a previous post, but if journals were as efficient as blogs in finding and correcting mistakes my life would be a lot less stressful.

As far as the bit on “who is the best person to submit this to a journal” I’ve got an idea. I think Tango (et al. LOL) should be given credit for the work that would eventually end up in a publication. But I also agree with the sentiment that it’s probably not worth his time. So maybe the best solution is for one of these young economists to co-author a paper with Tango. That would seem to be the best outcome for all parties.

God do I wish “non-academics” gave this much of a damn about my work. It would mean virtually free publications in terms of effort.

Now, if you’ll excuse me I have to go finish a first draft of my latest work, so I can send it to my co-authors… and get their feedback, send it back to them, get approval for submission, submit it, get reviews back, make corrections, send the corrections to the co-authors, make their changes, send it back to them once more for re-submission approval, re-submit it, get a last set of changes from the review process, make THOSE changes, submit it one last time, see the article in “press format,” make editorial and blocking changes, send that to co-authors, and get their approval for final submission… and of course that’s all assuming the paper is accepted on first submission!

Sal/27:
“Unfortunately, it sounds like the academic community isn’t anxious to embrace blogs and amateurs as a complement to peer review. And that’s a pity.”

I think you’re bound to find mixed feelings on this statement. I wouldn’t generalize in this way; the early career generation of academics seem to be fed up with the publishing industry. At least in my field, they charge us “printing fees” just to have the right for our work to appear in their journal, and then also charge our academic institutions for access to our own work. It feels at times they they’re just leeches making a living off our hard work. That’s probably too harsh, but it is how it makes one feel at times.

Given the publishing tools available to bloggers, the instant feedback, and the low entry level for providing feedback, more and more academics are seeing the value in putting their work online. Many online journals have already cropped up. I think it’s only a matter of time until we see “primary, peer-reviewed research” published without journals at all, but instead on something very much like a blog.

Basically, the journals are going to be in trouble for the same reason the newspaper industry is. In fact, I think the problem is worse for the journals. The only thing keeping them in business is the “score keeping” for promotion and tenure in academic circles. If not for that, I’d suspect submission rates to journals would drop dramatically.

To the point about online blogs and academics.  I do think there are some that embrace the idea of using the internet as a platform to get things across.  I do think that journals often charge a bit much for their publications.  Authors like Rod Fort have almost all of their work published online and have embraced sharing data and information.  Bradbury and Berri as well must believe that the online forum is a great tool given that they post so often.  I’m not sure the copyright laws on allowing your own work to be seen by anyone, but I think it’s good practice.  The reason I often refer people over to that site is because of this.

I think Shawngoldman makes some good points, which Colin also made over on my site.  I really think that in order for both sides to understand each other and be progressive about things, there needs to be easier access to all work.  At the same time, critiquing the work comes with a responsibility, and answering every person that questions something in a study would be tiresome for authors as well.  I think there are always going to be things pointed out as flaws in a study, because there really aren’t any perfect ones.  And I don’t think anyone here, Tango, Colin, Guy, would argue that there are always shortcomings with statistical analysis.

That said, I’m curious how something like this would be set up?  Just a curiosity.

Trying to prove that Moneyball increased the price of OBP is a fool’s errand.  The only place you can really look is the free agent market, and there just aren’t that many free agents every year to give you enough of a sample.  The player salaries are affected by other things like position, defensive ability, age, scouting opinions.  Some of this information isn’t even available.  You can control for the rest with regression, but I don’t think any study could sift through all these factors, given the small sample, and come up with any conclusion that falls outside an error range.

Rally makes a pertinent point here.

Look at what we found out with the data from Saber Rattling: that the WAR salary model worked fairly well… EXCEPT for those players who signed a 1-yr deal.  Those guys got discounted by 45%.

I would say that there is no (non-baseball-fan) academician that would have told us that.  A passing baseball-fan academician might have figured that out.

Half the true baseball fan, given that data, would have been able to tell you that 1-yr deals were being discounted (maybe not by telling you by how much, but certainly telling you that 1-yr deals were being discounted).

So, when you are trying to find an understanding in the labor market, it’s the baseball fan who can tell you more.  Ideally, you’d like your academician to be a baseball fan as well.  But even then, it seems the academic brain takes over.

What would the Sakes/Hauer model tell us about what happened in 2009?  Well, seeing that their parameters did not include the number of years that each player signed for, then they’d be missing the ENTIRE story of the 2009 pre-season.  And so, what happens with their model?  Well, all the other parameters, whichever ones would more closely correlate to the missing parameter (number of years of contract) would be OVERWEIGHTED.

In the perfect correlation is not causation, those parameters will act as proxy, in part, to the missing parameter.  But, the academician, believing that he’s caught all the major parameters, will not even fathom the idea that they are missing the most important parameter in their model, and will therefore conclude that correlation is causation (in that case).

At the same time, critiquing the work comes with a responsibility, and answering every person that questions something in a study would be tiresome for authors as well.

Bloggers have no problem here.  There are so far 38,000 comments in this blog, and MGL and I account for 14,500 of them, which is about 10 blog posts a day between the two of us, over and above the threads we start.

Hence the proclamation that you are a “prolific writer” by Bradbury.  My comment wasn’t meant as a criticism toward blogging, but a definte hurdle for those in an academic world.  If I didn’t find value in publicly presenting ideas, I wouldn’t do it. 

This is especially true when an academic reputation is on the line, and ensuring careful retorts takes significant amounts of time.  I also think there is a slippery slope (god I hate that term) in which the professional courtesy taken through the review process would be diminished. 

With lots of internet opinions, it is much more difficult to get a point across in a cordial manner.  And the last thing academic researchers want is for everyone to be bashing each others’ work on a public forum.  I also think this is where the worry of accountability with pseudonyms (which I proposed weren’t a huge deal at sites like this, but more so at BBTF and the likes) takes effect.

I know that, on my blog, I use it to jot down ideas and hopefully spark new ones for myself.  I would absolutely love to post things there that have to do with work I do as a student researcher, but unfortunately the incentives skew me toward journal publication so I can maybe get a job one day in the field.  Presenting that work on a blog would probably not be optimal for my own future.  But I try to do other fun, hobby-like discussion when I can.  It’s just that the best use of time for myself is on other things. 

This DOES NOT mean that the best use of your time or other interested in baseball analysis is to become a college professor or no longer using the internet to spread ideas, OR that it’s significantly less important to everyone else.

I think the same argument could be made for any academic.  It’s not that it isn’t a worthwhile use of time intellectually, but it becomes difficult to dedicate significant portions of time to internet discussion, when a University expects them to publish.  The same goes for you and the cohort here: the advantage to submitting to journals is minimal and not a best use of time.  I think this is one of the HUGE hurdles that anyone faces when trying to reconcile the two disciplines.

The first thing those in the academic world should want is the truth. The source frankly shouldn’t matter if the argument is well formulated.

Now I think it’s fair for an economist to side with credentials on matters of opinion-stat heads often evoke similar rhetoric when weighting opinions.

That said, the last thing those in the academic world want is to retract a paper or have it rendered irrelevant.

There’s a lot of conflicts of interest in academia that can get get in the way of truth. It’s understandable why a field might be resistant to a “voice from the wilderness”.

It’s a shame though.

Millsy notes on his site that a forthcoming special issue of JSE edited by Rodney Fort will feature an article where “Berri and Bradbury relate to us their extensive experience with the SABR-metric and APBR-metric communities, named after the Society for American Baseball Research and the Association for Professional Basketball Research, respectively. The main issues are 1) the
interplay between these informal communities and the rest of us and 2) that the rest of us live
according to the formal peer review process.” See Fort’s introduction to the issue here:  http://www.rodneyfort.com/Academic/CurrentEfforts/Forthcoming/FortJSE.pdf. 

I have to say, Millsy, that if Fort is interested in fostering a productive relationship between the academic and amateur camps, commissioning this article was probably a very bad idea.  These are literally the two economists in America with the worst relationships with these two analytic communities.  Berri, in particular, has basically been at war with the APBR community.  Even if you think that’s largely the fault of the amateurs, it seems odd to think Berri is the guy to help economists navigate these waters.  I will be shocked if this article does not serve to drive a larger wedge between the academic and non-academic worlds (hope I’m wrong, though).

Guy/35,

I hope you’re wrong as well.  However, while I think that Bradbury and Berri are probably the most opinionated, the fact that they have immersed themselves in the medium is the basis for their expertise in the article.  I think having that perspective is beneficial.

I can’t comment much further than that.  I would hope that in an academic journal, fuel for discussion is presented, rather than backlash.  I hope that anyone reading the article first sits and thinks about the points, before becoming combatitive.  Given the discussion here, I have plenty of confidence that would happen.

I will do my best to withhold judgment until the article is published.  But I would say that only Bradbury has “immersed” himself in the non-academic analytics community; neither Berri nor anyone in the APBR community would say he has had anything to do with them. 

On post #33:  As Millsy says, young academics face requirements that would preclude them from devoting a lot of time to online commentary.  But that’s not the only option.  They can follow the most important online discussions w/o a huge time commitment, and learn from them.  And most importantly, they can use subject matter experts to review academic papers prior to submitting them for publication.  Tango makes himself available for that, I’m sure Phil Birnbaum would too, I’ve done it in a few cases.  We won’t be able (usually) to tell you if you’ve cited the right prior work in your field, or used the best possible econometric techniques—but you can get that elsewhere.  We can often tell you of relevant sabermetric work, and—frankly—will often just be better at finding flaws in your work and suggesting better approaches.  It amazes me this isn’t done more often than it is.

One other point I would make is that while the most obvious contribution of the amateur analysts is subject matter expertise, there are also many cases where economists just fail to do good science in terms of seeing problems of sample bias, or putting the magnitude of the relationships they study in the proper perspective.  In some cases, like the Hakes-Sauer paper or JC’s ill-fated valuation model, the saber community is even doing better economics! (And this is where, understandably, it’s hardest for the economists to handle criticism.)

In some cases, like the Hakes-Sauer paper or JC’s ill-fated valuation model, the saber community is even doing better economics! (And this is where, understandably, it’s hardest for the economists to handle criticism.)

Guy is exactly right.

The entire baseball model is indeed just like the Fantasy baseball model.  So, there are literally millions of people who are better versed in the economic model of MLB payroll than JC, among others.  It’s hard for them to accept, but there it stands.

Rodney Fort said it best at Phil’s blog once when he said that economic theory is just that, theory, until it is tested against reality.  And it’s the WAR model that corresponds to reality, which itself is a precursor to the Fantasy baseball model.

And Berri does not interact with the analytic community.  At all.  And JC, while maybe keeping abreast, maybe, does not interact with the analytic community either.

When someone has a minute, can I get some more low-protein gruel?

Guy/37,

Thanks for the response.  I agree it’s not the only option.  But it is the one that could possibly foster the most interaction.  However, I do follow as many discussions online as possible.  I try to comment when I can, but often it’s not as detailed as I would like to make a full refutation, which is where I think there is a disconnect.

Unfortunately, much of the online stuff is unrelated to the type of things I’m interested in with my own research.  That’s not to say I don’t find it interesting, to the contrary, but my professional sports interest lies elsewhere.  That doesn’t mean that people outside my ‘lab’ can’t be useful.

I think that sometimes the “failure to do good science” is confused with the choices that economists often must make, or are used to making with their data.  Much of the field of econometrics is finding ways in which to minimize bias, given that it’s not usually an experimental sample.  Some call it jumping through mathematical hoops.  Others call it an art.  Some of the methods I’ve seen are so mind-blowingly complicated I wonder if it’s really worth it.  But completely avoiding bias is almost impossible with much of this type of data, and deciding to do one thing will ultimately cost the researcher in some way in another area of the methodology.  I think many of the prominent saberists completely understand this, but I worry that others may not.

When it comes to this, I’ve found that economists are far superior to many other social scientists that do not work with experimental data.  At the same time, the success in that realm has left a divide in how different disciplines think or communicate about problems...which hasn’t necessarily been a good thing.  Then, yes, there are plenty who do ignore methodological problems altogether...but this is not a problem specific to economics.  I would argue it’s more specific to other non-experimental disciplines, albeit without as much ‘confidence’.

The point is well taken that sabermetricians have had much to offer to the study of baseball.  And perhaps some of the research has done work refuting some economic models.  As a further point, let’s not go as far as to proclaim sabermetricians better economists than economists themselves. 

Whether you think so, whether its true, whether it’s absurd...saying so is going to do the exact opposite that many here have said they hope can happen: using one another to leverage research among both communities.  To deny that this is hubris is being dishonest.  While saying that this is an overwhelming theme likely exagerrates things, I’m not sure this disproves any point that I had made in my original post. 

I am NOT making this point in order to be difficult...but in hope to shed light on where the possible huffiness can come from.  And I admit that this can happen on either side of the spectrum.

Here’s the view from where I sit.

Econometricians and sabermetricians differ in one key respect. Econometricians study economic principles and econometric tools and look to apply them to the study of baseball. Sabermetricians study baseball, and turn to econometrics as one (but not the principle or only) way to conduct that study.

What’s really staggering these days is the amount of cheap, readily-available computing power that we have. A person with an inexpensive personal computer can download something like R or gretl for free and gain access to a lot of sophisticated econometric techniques. I think we’re a little smarter about using them than some academic econometricians give us credit for, but certainly we’re not as sophisticated as the full-time academics are about them. (Bradbury mentioned a couple of books the other day, and I’m going to try to pick them up when I can.)

On the other hand, from the perspective of a sabermetrician, econometricians seem to be ruled by the law of the instrument - “When the only tool you have is a hammer, it is tempting to treat everything as if it were a nail.” The econometrician’s hammer is regression analysis - certainly a powerful tool, but even the best hammer is a poor wrench. Most econometric measures of “goodness of fit” are painfully unable to tell the difference between Basic RC and a sophisticated version of BaseRuns, for instance.

To compound the issue, there are numerous barriers to entry for an amature wishing to engage with the academic study of baseball - journals are expensive and most of the good methods of searching for academic papers aren’t available to people who aren’t at an academic institution. The costs - both monetary and otherwise - of publishing in a peer-reviewed context are greater for someone without an academic affiliation.

And some academics in the economics field treat those barriers to entry like membership in some sort of priesthood or medieval artisan’s guild. Bradbury, for instance, acts as though its patently impossible for laypeople to find an error in an academic paper that has been peer reviewed. (And if that’s an inaccurate reading of Bradbury, then Bradbury is simply a poor writer.)

If we’re looking to pinpoint where communication breaks down between academic researchers who study baseball and sabermetricians, I think any fair appraisal has to acknowlege that a lot of it has to do with the wall that academia has build around itself.

Colin,

As I’ve said before, I think those are fair points when it comes to analysis.  I, for one, have attempted to also surround myself with those academics versed in other statistical techniques (whether that be FA, PCA, cluster analysis, among other topics).  And I also support that there are plenty of smart people that can talk about academic works.

On another note, I like that you bring up R here.  I think it’s the most powerful tool out there.

"As a further point, let’s not go as far as to proclaim sabermetricians better economists than economists themselves.  “

No one is making that claim.  At least with payroll models, at the least, are sabermetricians providing better clarity than economists.  Whatever that means, that’s what it is.

That doesn’t mean you get to extrapolate that statement outside that data point. 

See Fort’s introduction to the issue here:  http://www.rodneyfort.com/Academic/CurrentEfforts/Forthcoming/FortJSE.pdf.

Here’s the description:

Berri and Bradbury relate to us their extensive experience with the SABR-metric and APBR-metric communities, named after the Society for American Baseball Research and the Association for Professional Basketball Research, respectively. The main issues are 1) the interplay between these informal communities and the rest of us and 2) that the rest of us live according to the formal peer review process. The authors provide both encouragements for interaction as well as warnings about the pitfalls of this type of interaction. There are tensions and jealousies aroused over the issue of proper credential and peer review. There is the issue of proper assignation of credit; when incorporating a technique or measurement developed outside the peer review process, what to do? Finally, there is the issue of dealing on line with these fellow travelers who often adopt the anonymity of pseudonyms. Dave and JC offer an array of examples to make their point, a solid discussion of how to evaluate different measures from the economic perspective, and a nice history lesson on the evolution of some important measurement devices commonly used in sports economics.

I have to say that doesn’t make me optimistic that JC is dialing back his antagonism toward the sabermetric community.

On another note, I like that you bring up R here.  I think it’s the most powerful tool out there.

Off topic but I’m interested in what tools people are using for their sabremetrics.  I started playing around with jmp a couple of days ago and I’m having a good time with it.  Any recommendations?  (note: I’m probably not serious enough to learn R at this point.)

It is ridiculous to appeal to the authority of a PhD in an unrelated field - my graduate degree in EE is worthless in discussing baseball.  The irony is that within academic circles, you’d get laughed at for flaunting your credentials as an academic at a third-rate institution like Southern Utah University (Berri) or Kennesaw State University (Bradbury). 

Berri’s response to Roland Beech’s review was particularly poor - Roland pointed out that Berri’s NBA metrics were obsolete; Berri responded by claiming that Roland was mad because he (Berri) wasn’t using Roland’s statistics in his book.  Which was untrue, of course.  Roland’s a guy who puts his money where his mouth is by working in the NBA; Berri, from looking at his CV, has not done so. 

Analytical jobs are a very small subset of total sports employment, but Tango and MGL and likely many others here here have worked for various pro teams at one time or another.  It stands to reason that if academic economists had all the right answers, they would be working for professional teams and pseudonymous bloggers or professional gamblers with nothing more than a love of sports and some math and programming talent would not.  (Most people who work on the operations side of pro teams who weren’t pro athletes are lawyers, not economists.)

Cross @ 46: Matlab would be my preferred way to go.  You can download web pages with it; you have access to Perl/Regex commands; it can handle huge arrays; and it does nice plotting.  And the language is basically a simple version of ‘C’.

Matlab also costs $1900.00 per seat.

I use MySQL for data processing and any number crunching I can code there (which is more than one might think, although admitedly I abuse the privelege a bit) and do further number crunching in any of Excel, gretl or R. gretl and R are both free - gretl is loads easier to use than R, even with R Commander installed. But R is the most powerful of the lot. I do some work in Python and am considering moving more in that direction with IPython/SciPy/NumPy.

The only stuff I really work with is Pitch f/x, and I just need SQL and Excel for that.  I’ve used R a couple of times, but it’s really hard to use and doesn’t really offer anything that you can’t get with Excel add-ons - or at least nothing that I need.

Well, academic matlab is $95 and you can buy it and use it if you are a student, teacher or parent or work at a non-profit.  So the vast majority of people here don’t need to pay $1900.

XL Stat is also pretty good - it only costs 50 bucks if you are a student.

Thanks for all the info guys!

I really think R, once hitting the learning curve, has a lot to offer.  There are a number of pre-developed packages at their open source area for a lot of differnet things.

As for visualization, sometimes R itself can fall short unless you really know what you’re doing.  I’m not one that can do great visualizations in R, but I do know it can be combined with other visualization programs to improve its capabilities.  If you’re looking for ease, Excel and SPSS are the way to go.  However, they have extreme limitations (especially SPSS).  R is probably a little easier than Matlab given the open source programs available...Matlab seems to be more of a ground up matrix manipulation program that does need fairly significant programming or mathematical knowledge.  I learned through R here at school, so that is the most user friendly to me.  It’s really a matter of preference and what you originally learned on.  I’ve found myself having trouble using some of the language and programs in GAUSS as it turns out.  You can also get a free version of GAUSS through the command line (at least in Windows).  That is limited to 100x100 matrices (10,000 cells), so it depends on what you’re doing.

Tango,

I don’t think I extrapolated THAT far from this:

“In some cases, like the Hakes-Sauer paper or JC’s ill-fated valuation model, the saber community is even doing better economics!”

I think there are some advantages to the way things are explained in the sabermetric community in terms of analyses.  The comments here have been mostly very productive, and very much so from Guy.  I was simply pointing out where miscommunication could take place, as some could definitely take it as a loaded statement, even if that is not its intent.

I don’t know how good it is for statistics/econometrics, but Octave is a free/open source replacement for Matlab, at least for Linux.

Millsy:  I meant my comment to convey that it is and should be surprising that economists need to learn anything whatsoever from us about the economic issues.  And I would never claim the amateurs are “better economists” in any general sense.  But for some reason, guys like Tango and Studes have a better handle on this salary market than the credentialed economists.  And it’s not just saber “experts”—as I’ve noted before, every fantasy baseball player in America knows that the last few players on a team only cost the minimum salary—so I don’t know why replacement remains a tough concept for some economists (but we don’t need to litigate that issue again). 

Indeed, one of my frustrations is that the economists who work in this area haven’t been able to advance our understanding of player valuation.  For example, we’ve struggled here with how to reconcile the fact of one labor market, in which wins cost the same for every team, with the reality of hugely varying MRPs by team.  If the Pirates and Red Sox must pay the same, but the Sox earn far more revenue from any given unit of performance, what are the resulting prices?  Is it the average MRP of all teams?  Something else?  If it’s the average, that means half the teams are losing money each time they hire a free agent—so why do they ever do that?  (Maybe only as often as necessary to keep the luxury tax gravy train alive.) These are interesting issues I which a good labor economist would tackle.  But I haven’t seen it.  (If you know of such work, clue us in.)

Guy,

I understand the reason for the comment.  I think your comment was reasonable.  I was simply pointing out where misunderstading can definitely take place.

As for getting pure labor economists involved in the baseball labor market, it’s something I’m very interested in seeing as well.  I’m not sure any rational economist would argue that valuations of players are vastly different for each team though.  In fact, I know this is one of the main points of any sports economics class I have come across.  I actually had asked Bradbury in the past about this aspect of his model, which he stated that they are in fact averages across MLB.  I think averages coming out of a model are okay in terms of estimation of what a player would get, as long as we understand what they are.  But I think it’s also important to take into account the way the talent gets dispersed.

Looking at the labor market and how talent really gets dispersed is something I’ve been curious about in research, but don’t really have the tools myself to fully analyze.  I haven’t fully immersed myself into labor econ much further than the master’s level, and I’m hesitant to feel I have enough grasp on the subject to comment further there.  I know Fort has a new paper posted on his site with Jason Winfree discussing the supply of talent in closed leagues.  It’s very much on the theoretical side of Econ, but calls for empirical work in the area.  I’d link it here, but there seems to be an error on the site.  Here’s the citation:

Fort, R. & Winfree, J. (2009).  Sports really are different: The contest success function and the supply of talent.  Review of Industrial Organization, 34, 69-80.

Honestly, it’s a bit dense.  To fully digest, it took me about 3 or 4 reads.

Fort and Quirk also have 2 papers forthcoming that discuss balance in season and single game ticket leagues.  I think one of the interesting findings here is that the response to winning is different in different cities (a point which we would expect, of course).  The interesting tidbit here is that, given this, it’s likely better that a team like the Yankees win more often from a total welfare standpoint.  Again this work is very theoretical and calls for more empirical work.  And again, a few careful reads are required to get everything put out.  While these don’t get into the nitty gritty of analyzing the labor market, I think they DO lay a foundation for your curiosity in the subject.  Both of these papers are available on the webite under the listing of forthcoming papers.

To post a link, just post it as a straight text like this:

http://www.tangotiger.net

Sorry, I guess I wasn’t clear.  For some reason the file won’t load when I’m at the site.  Here’s the link to the article list:

http://rodneyfort.com/Academic/Academic.html

I don’t see that paper listed at all. I do see this paper from the same authors:

“Fixed Supply, Nash Conjectures, and Sports League Modeling.”

"I’m not sure any rational economist would argue that valuations of players are vastly different for each team though.”

Millsy, I’m not sure what you meant by this.  I was saying that teams generate very different amount of revenue from X amount of player talent (i.e. wins), based both on size of market and where on the win curve they are (a team’s 92nd win is worth much more than the 62nd).  These are not controversial ideas within the saber community. 

Nonetheless, teams all pay the same amount for X amount of talent (except maybe the Yankees).

Are you saying sports economists would disagree with one or both of these forumlations?

No no no.  Sorry, perhaps the response is confusing.  I think the price of talent is generally fixed in leagues--though teams like the Lakers and Yankees may have some differences.  And I agree with the ‘sweet spot’ argument without a doubt.  I think I was agreeing with the fact that the revenue produced across markets for wins differs considerably.  It looks like I forgot to include a few words in that sentence, sorry about that. 

Colin,

The Winfree/Fort paper should be toward the top (#84 on his site).  The other two are #85 and #86.  They’re listed as forthcoming in JSE and Economic Inquiry, so they’re just manuscripts (and should be treated as rough drafts).

Colin/#60, the list of articles is under his Vita page.

I found a link to the paper at this site:
http://www.springerlink.com/content/h753276p70725k47/fulltext.pdf

Thanks, Mike.  It’s one I had seen a presentation on, but wanted to read again.

most of the good methods of searching for academic papers aren’t available to people who aren’t at an academic institution

I have found that the best freely available tool for searching academic papers is Google Scholar. 

It often returns the most pertinent papers and searches a very wide pool of journals.  In addition, it gives the articles which cite a give article which is helpful.

Guy,

I’ve been thinking about this a little more:

“For example, we’ve struggled here with how to reconcile the fact of one labor market, in which wins cost the same for every team, with the reality of hugely varying MRPs by team.  If the Pirates and Red Sox must pay the same, but the Sox earn far more revenue from any given unit of performance, what are the resulting prices?  Is it the average MRP of all teams?  Something else?  If it’s the average, that means half the teams are losing money each time they hire a free agent—so why do they ever do that?”

I’m trying to recall exactly how the discussion went when learning some things in my sports economics class.  I believe the argument goes, in general, the idea is a fixed price for talent.  The theory is that the teams choose the talent level as a long-run choice.  Therefore, given a fixed price of talent, the payroll is fixed at X for each team to maximize profts.  Let’s say the Yankees feel they maximize profits at 94 wins based on the decreasing returns to adding more talent, while the Pirates feel the sum of purchased talent is optimal at 60 wins (subject to whatever uncertainty).  Given different responses to winning, the total amount of talent on a team is chosen individually.  Rather than try to buy more wins, the Pirates choose 60 wins as the way they maximize profits, given the market price of the talent.  Perhaps this is where an economist would start from 0, while others start from the replacement level.  We can call “talent” whatever we like here.

At the going rate of, say, $1 million in revenue per win (from 0, not replacement just to make things extremely simple).  The Pirates buy a combination of players, maybe 1 superstar, that provides them with 60 wins.  Meanwhile, the Yankees buy 6 superstars to get to 97 wins, with the rest adding some marginal number each.  We’ll assume the 61st and 98th win in each market give the team only $999,999 in revenue, so there’s no incentive to buy any more talent for either team (again, working in simplistic whole numbers).  They choose this in the long run, rather than change it each year based on what they have (of course, I’m sure this happens to a certain extent depending on the status of minor league talent, etc.). 

Could this be a possible disconnect between saber valuations vs. others?  It relates back to the market size, as well as the elasticity of demand in each market (which could effect the actual relevance of the sweet spot), and therefore keeps any team like the Pirates from buying enough talent at its own market MRP to compete above the 60 win mark.  The market price for talent, I guess, works itself out so that it’s fairly static, given the fluid-ish nature of free agents coming and going.

Taking the two-team league for example, with 7 superstars in the market.  The Yanks buy the 6 superstars getting them to, say, 84 wins (14 each--I’m not bothering with 25 man rosters or anything near it for this, but further talent buying should follow similarly), so that the 7th at market price would put them above what they’d be willing to pay.  There is 1 superstar left on the market, who goes to the Pirates, because they could use the 14 win player.  The Pirates are willing to pay a total of $60 million on payroll, so 14 wins is worth buying.  However, the Yankees won’t let him go for less than they would pay for a 13 win player (which would get them to 97), the player goes at the $13 million-ish price.  From there, the same thing happens for the rest of the talent in the league at a similarly pushed up price by the Yankees or whatever larger market team is in the mix.  There’s a slight difference in payment, but not one we would find in a more fluid market with 30 teams, hence the price for talent generally looking the same across the league. 

It wouldn’t be that teams are losing money on this talent, but that they are paying for different sums of talent up to what their demand function allows.

Now I can’t say if all of those assumptions hold.  I’m just trying to lay out a story here that might make sense.  I’m sure some of it could be refuted.  I definitely think that Tango would have some interesting insight into how the front office does its budget in terms of the investment allowed in talent.  I don’t know how much information they would allow him to share, or if his duties are more along the lines of just telling them who the best players are.  But I do think it would be enlightening.

I made a mistake above.  I meant to say the going rate to pay is “$1 million per win”, not the revenue.  The calculus of the problme is probably oversimplified, but I think it shows where the marginal value of a player could be higher for the Pirates than the Yankees at some point in time.

There’s been several references in this thread to those in the online community as ‘amateurs’.

It reminded me of a conversation I had this summer. There’s an MLB front office guy that I’ve made friends with, who incidentally has a degree in econometrics. While discussing all things baseball while taking in a game, I also referred to us as ‘amateurs’ and he cut me off, taking exception to that characterization, expressing his opinion that some of the best work being done is in our community. And when I mentioned a point I made on this blog, he said, “I saw that”

I don’t think that there is any doubt (based on the amount of jobs given to members of this community and others like it) that the baseball front offices are watching and making the best of the best into pros.

I do think that people like JC think that we are amateurs in comparison to their (lofty) academic credentials. And that’s where lots of the disconnect and dislike is coming from I’d guess—the huge gap in evaluation. Like Francouer being worth millions for going out there every day despite being worse than freely available talent.

“This can also explain the conflict between those who extol peer review, despite its many flaws, and downplay open source science. They are controlling view scientists protecting their turf and power and prerogatives. Anyone thinking about the ideals of science, the classical view of science, immediately realizes that open sourcing the arguments and data will meet the ends of extending knowledge much better than peer review, now that it is possible. Peer review was a stop gap means of getting a quick review of a paper that was necessary when the costs of distributing information was high, but it is now obsolescent at best.

“Instead the senior scientists and journal editors are protecting their power by protecting peer review.

“Bureaucrats, and especially teachers, will tend strongly toward the signaling and control view.”

Whole thing here.

Excellent point.

The question boils down to this: suppose we never had these academic peer reviews.  And suppose we are where we are in terms of the information super highway.  Feedback is instantenous.  Credentials are applied by other feedbackers (like an Ebay or Amazon commenting/review system).  A guy like Phil Birnbaum would bubble up to the top, for example, as other reviewers review the reviewers, giving him more clout, for example.  The whole thing just starts to shake out like a Wikipedia, or even Google-like system.

Given this existence that we are in today, someone comes along and says “we academics need to have a peer review process”.  Will it see the light of day?  Who would sign on to such a venture?  And, would it be accepted as the best process to review work?